Imp (& agent) becoming very unstable

Agent Code
1

Don’t know what happened. I made an error sending a blob over the agent.send() method, which on the agent side made things go wrong. Since then the whole setup is unstable . Log:

Sat Mar 23 2013 15:25:34 GMT+0100 (Romance (standaardtijd)): Device booting
Sat Mar 23 2013 15:25:34 GMT+0100 (Romance (standaardtijd)): Device configured to be "PoolControl"
Sat Mar 23 2013 15:25:34 GMT+0100 (Romance (standaardtijd)): IMP : Started OK
Sat Mar 23 2013 15:25:49 GMT+0100 (Romance (standaardtijd)): Power state: online=>offline
Sat Mar 23 2013 15:25:55 GMT+0100 (Romance (standaardtijd)): Power state: offline=>online
Sat Mar 23 2013 15:26:45 GMT+0100 (Romance (standaardtijd)): Power state: online=>offline

Code used to run fine until now…

Just tried with some old test code which worked perfectly => same behaviour. I assume something is wrong with the infrastructure

2

Would agree can’t reach my Agents at the moment

3

Seems to be general behaviour:

  • Imp downloads code & starts
  • Agent is not running
  • Imp goes offline after 10-15sec
  • Imp goes back online but doesn’t start up
  • imp goes back offline after about a minute and stays offline…
4

Same here.

5

Things should be running now, but yes the agents have been upset this morning.

@vedecoid will contact you to get the code that caused this, looks like you have a repeatable bug :slight_smile:

6

that’s what beta programs are for :slight_smile:

7

The fix is now live, it did appear to be related to trying to server.log() a blob.

Thanks for posting about the bug - if you ever do manage to seemingly crash the agent VM, info about what you were doing is most useful for debugging (and writing new unit tests!)

8

I’ve got the same problem right now, device.on() function doesn’t get triggered in the agent when issueing agent.send() on the imp. any suggestions?

9

Still working on tracking down what happened this time (but it’s a different bug), but the auto-restart didn’t kick in for a likely totally unrelated reason which exacerbated the issue. Working on that too!